CVE-2026-81094: mcp-router CLI before 0.6.3 Binds the MCP Aggregator to All Interfaces Without Requiring Authentication
The mcp-router CLI served its MCP aggregator on every interface and enforced authentication only when the operator asked for it. The serve command in apps/cli/src/commands/serve.ts defaulted its host to the all-interfaces address on a fixed port, and required a token only when the corresponding flag was supplied, so a default invocation exposed the aggregator, and every MCP server it fronted, to anyone able to reach the port. Release 0.6.3 defaults the host to the loopback address and refuses to start without a token whenever the host it is given is not a loopback address; no earlier release carries either check.
Security readout for executives and security teams
Plain-English summary
mcp-router CLI versions before 0.6.3 can unintentionally expose an MCP aggregator and its connected MCP servers over the network without authentication. A default launch may therefore allow any network-reachable party to access sensitive MCP capabilities or data. Version 0.6.3 changes the safe defaults and adds an authentication safeguard for non-loopback hosting.
Executive priority
Prioritize remediation immediately for internet-facing, cloud-hosted, shared-network, or otherwise untrusted-reachable instances. Upgrade all remaining pre-0.6.3 deployments promptly because default operation is unsafe and may expose downstream MCP services. Investigate reachable historical deployments for unauthorized access, while recognizing that the supplied evidence does not confirm exploitation.
Technical view
The serve command defaulted to an all-interfaces address on a fixed port, while token authentication was optional. This creates a network-accessible, unauthenticated endpoint under default settings. CVE-2026-81094 is rated CVSS 4.0 9.3 and CWE-306. Release 0.6.3 defaults to loopback and rejects non-loopback hosting without a token.
Likely exposure
Highest exposure exists where mcp-router CLI before 0.6.3 runs with default settings and the listening port is reachable from untrusted networks. Risk also extends to every MCP server fronted by that aggregator. Loopback-only or effectively firewalled instances have lower practical exposure, but still require configuration and version verification.
Exploitation context
The supplied sources establish low-complexity, unauthenticated network access with no user interaction. They do not establish observed exploitation, and the record is not identified as being in KEV. Treat internet-accessible or broadly reachable instances as urgent exposure without representing exploitation as confirmed.
Researcher notes
The vulnerable behavior is the combination of all-interface default binding and optional authentication. The confidentiality and integrity impacts are rated high; availability impact is not claimed in the supplied vector. Exposure depends on routing, firewalling, bind configuration, token use, and downstream MCP capabilities. Earlier releases lack both safeguards introduced in 0.6.3.
Mitigation direction
Upgrade mcp-router CLI to version 0.6.3 or later.
Bind the aggregator to loopback unless remote access is explicitly required.
Require a strong token whenever hosting on a non-loopback address.
Restrict aggregator access through network controls to trusted systems only.
Review current vendor guidance before relying on compensating controls.
Validation and detection
Inventory all mcp-router CLI deployments and record their installed versions.
Confirm each aggregator's effective listening address and network reachability.
Verify non-loopback instances cannot start or operate without token authentication.
Identify every MCP server exposed through each affected aggregator.
Review network telemetry for unexpected connections to previously reachable aggregators.
Generated from the cited source records. This long-tail analysis has not been individually reviewed by a named human.
Potential ATT&CK relevance
Conservative CVE-to-ATT&CK context
These mappings and lookup hints may be relevant to the vulnerability behavior, CWE, affected product, or exposure path. Glexia-inferred context is not an official MITRE, ATT&CK, CWE, or CVE Program mapping.
ATT&CK lookup starting points
Use these exact CWE pages and searches to review the Glexia ATT&CK library from this CVE's weakness and description context.
cwe · medium confidence lookup
CWE-306: Credential and account abuse lookup
Authentication and credential weaknesses can make valid-account abuse and credential telemetry useful review starting points. Open the exact CWE lookup page first, then review the ATT&CK searches from that MITRE weakness context. This is a Glexia lookup hint, not an official ATT&CK mapping.
The CVE wording references authentication or credential exposure, so valid-account and credential-access review may help. This is a Glexia inferred lookup path, not an official MITRE, ATT&CK, or CVE Program mapping.
These fields come from the CVE record and ADP containers, not from Glexia's Take. They preserve time-varying source decisions such as CISA SSVC, KEV status, CVSS metrics, and provider references.
2CVSS vectors
3Timeline events
1ADP providers
6Source links
SSVC decision data
CISA-ADPCISA Coordinator
Timestamp
Version
2.0.3
Exploitation: noneAutomatable: yesTechnical Impact: total
CVSS vector scores
2 official scores
We collect every scored CVSS vector available in the official CNA and ADP containers. When more than one version is present, the table keeps the source vectors side by side instead of collapsing them into the highest score.